Fujitsu takes SAW-less transceiver to LTE

Some time ago the wireless transceiver team at Fujitsu-mainly composed of Motorola/Freescale veterans-created a significant stir in the handset world with a transceiver IC that required no inter-stage SAW filters. The basic design idea was that if you don’t generate energy at spurious frequencies in the first place, you don’t need inter-stage filters to clean it up. So the Fujitsu team developed a combination of low-noise circuit techniques and low-noise DSP algorithms that together successfully eliminated the need for inter-stage SAW filters from most handset designs. This was a particularly valuable achievement in multi-mode handsets, where the number of SAW packages could get out of hand.

Now the same team is back with another achievement: a multi-band transceiver IC that covers GSM/EDGE, UTMS, and both the LTE modes-Frequency-Division Duplexing (FDD) and, primarily for China, Time-Division Duplexing (TDD). Once again, the chip requires no inter-stage SAW filters.

The challenges of bringing LTE into the architecture were numerous, according to Fujitsu senior director of RF engineering Vivek Bhan. First, LTE demands 20 Mb/s bandwidth in the receive chain, compared to less than 4 Mb for 3G standards, and also greater dynamic range than 3G requires. “We had to do new data converters from scratch with a new topology,” Bhan said. Second, because of the tight band spacing, the emissions restrictions for LTE are significantly tighter. This requirement caused, among other tasks, modifications to the chip’s DSP algorithms and hardware.

Bhan said the team started with a link budget and gain analysis on a spreadsheet. They moved from there to floating-point algorithm development in Matlab, then to a Signal-Processing Worksystem model, and only then to recoding the chip’s DSP engine. Important input during this process came from the carriers, Bhan said, who were performing similar exercises at the network level to understand just what they would need from handsets.

Another major issue with LTE is data bandwidth. “LTE is much more than a cell-phone standard,” Bhan said. “It also aims to replace wireline connections in areas such as broadband access and gaming. Category-3 LTE requires 100 Mb data rates. In most situations that speed can only come from 2x MIMO. So the chip has to have a second, independent receiver chain.”

By moving the LNAs on-chip and eliminating the inter-stage SAW filters, the Fujitsu 3G transceiver significantly reduced the assembled/tested cost of a multi-band, multi-mode 3G handset. Now, with stringent demands from LTE and with bands continuing to proliferate, Bhan’s team is showing just how scalable their concepts are. Bhan said the new chip is currently available in engineering samples, with production in Q4 after a metal-mask spin to improve reliability and yield.